yt_dlp/extractor/sbs.py

   1 from .common import InfoExtractor
   2 from ..utils import (
   3     smuggle_url,
   4     ExtractorError,
   5 )
   6
   7
   8 class SBSIE(InfoExtractor):
   9     IE_DESC = 'sbs.com.au'
  10     _VALID_URL = r'''(?x)
  11         https?://(?:www\.)?sbs\.com\.au/(?:
  12             ondemand(?:
  13                 /video/(?:single/)?|
  14                 /movie/[^/]+/|
  15                 .*?\bplay=|/watch/
  16             )|news/(?:embeds/)?video/
  17         )(?P<id>[0-9]+)'''
  18
  19     _TESTS = [{
  20         # Original URL is handled by the generic IE which finds the iframe:
  21         # http://www.sbs.com.au/thefeed/blog/2014/08/21/dingo-conservation
  22         'url': 'http://www.sbs.com.au/ondemand/video/single/320403011771/?source=drupal&vertical=thefeed',
  23         'md5': '3150cf278965eeabb5b4cea1c963fe0a',
  24         'info_dict': {
  25             'id': '_rFBPRPO4pMR',
  26             'ext': 'mp4',
  27             'title': 'Dingo Conservation (The Feed)',
  28             'description': 'md5:f250a9856fca50d22dec0b5b8015f8a5',
  29             'thumbnail': r're:http://.*\.jpg',
  30             'duration': 308,
  31             'timestamp': 1408613220,
  32             'upload_date': '20140821',
  33             'uploader': 'SBSC',
  34         },
  35     }, {
  36         'url': 'http://www.sbs.com.au/ondemand/video/320403011771/Dingo-Conservation-The-Feed',
  37         'only_matching': True,
  38     }, {
  39         'url': 'http://www.sbs.com.au/news/video/471395907773/The-Feed-July-9',
  40         'only_matching': True,
  41     }, {
  42         'url': 'https://www.sbs.com.au/ondemand/?play=1836638787723',
  43         'only_matching': True,
  44     }, {
  45         'url': 'https://www.sbs.com.au/ondemand/program/inside-windsor-castle?play=1283505731842',
  46         'only_matching': True,
  47     }, {
  48         'url': 'https://www.sbs.com.au/news/embeds/video/1840778819866',
  49         'only_matching': True,
  50     }, {
  51         'url': 'https://www.sbs.com.au/ondemand/watch/1698704451971',
  52         'only_matching': True,
  53     }, {
  54         'url': 'https://www.sbs.com.au/ondemand/movie/coherence/1469404227931',
  55         'only_matching': True,
  56     }, {
  57         'note': 'Live stream',
  58         'url': 'https://www.sbs.com.au/ondemand/video/1726824003663/sbs-24x7-live-stream-nsw',
  59         'only_matching': True,
  60     }]
  61
  62     def _real_extract(self, url):
  63         video_id = self._match_id(url)
  64         player_params = self._download_json(
  65             'http://www.sbs.com.au/api/video_pdkvars/id/%s?form=json' % video_id, video_id)
  66
  67         error = player_params.get('error')
  68         if error:
  69             error_message = 'Sorry, The video you are looking for does not exist.'
  70             video_data = error.get('results') or {}
  71             error_code = error.get('errorCode')
  72             if error_code == 'ComingSoon':
  73                 error_message = '%s is not yet available.' % video_data.get('title', '')
  74             elif error_code in ('Forbidden', 'intranetAccessOnly'):
  75                 error_message = 'Sorry, This video cannot be accessed via this website'
  76             elif error_code == 'Expired':
  77                 error_message = 'Sorry, %s is no longer available.' % video_data.get('title', '')
  78             raise ExtractorError('%s said: %s' % (self.IE_NAME, error_message), expected=True)
  79
  80         urls = player_params['releaseUrls']
  81         theplatform_url = (urls.get('progressive') or urls.get('html')
  82                            or urls.get('standard') or player_params['relatedItemsURL'])
  83
  84         return {
  85             '_type': 'url_transparent',
  86             'ie_key': 'ThePlatform',
  87             'id': video_id,
  88             'url': smuggle_url(self._proto_relative_url(theplatform_url), {'force_smil_url': True}),
  89             'is_live': player_params.get('streamType') == 'live',
  90         }